Break point fails with concurrent.futures.ProcessPoolExecutor in vscode python

[DonJayamanne]

@machineCYC commented on Thu Feb 21 2019

Environment data

• VS Code version: 1.31.1
• OS and version: Ubuntu 18.04
• Python version (& distribution if applicable, e.g. Anaconda): 3.6.5
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): virtualenv
• Relevant/affected Python packages and their versions: python inner packages

Expected behaviour

launch.json configuration as follows:

"configurations": [
  　{
      　"name": "Python",  
       　"type": "python",  
       　"request": "launch",  
       　"program": "${file}",  
       　"console": "integratedTerminal",  
       　"cwd":"${workspaceFolder}/Code"  
  }]

test python code as follows:

import concurrent.futures

FIBS = [28, 10]
 
def fib(n):
    print(n)
    if n < 2:
        return 1
    return fib(n - 1) + fib(n - 2)
 
 
with concurrent.futures.ProcessPoolExecutor() as executor:
    executor.map(fib, FIBS)

Actual behaviour

Set break point in function fib. Must be stop.

But it does not.

Steps to reproduce:

１.Copy the provided configuration to launch.json
２.Copy the provided python code
３.Run debug mode in python code with launch.json

Logs

Output for Python in the Output panel (View→Output, change the drop-down the upper-right of the Output panel to Python)

$ cd /home/yenchen/Desktop/temp_test/Code ; env "PYTHONIOENCODING=UTF-8" "PYTHONUNBUFFERED=1" /home/yenchen/basic_venv/bin/python3.6 /home/yenchen/.vscode/extensions/ms-python.python-2019.1.0/pythonFiles/ptvsd_launcher.py --default --client --host localhost --port 37647 /home/yenchen/Desktop/temp_test/Code/main.py

Output from Console under the Developer Tools panel (toggle Developer Tools on under Help; turn on source maps to make any tracebacks be useful by running Enable source map support for extension debugging)

Not anything.

[karthiknadig]

@machineCYC This requires "subProcess": true in launch config to begin with.

It looks like it primarily depends on os.fork. This is not yet supported #943. so you have to switch multiprocessing to use spawn. Also, when using spawn, main module has to finish loading, so use this code to run:

import multiprocessing
multiprocessing.set_start_method('spawn', True)
import concurrent.futures

FIBS = [28, 10]

def fib(n):
    print(n)
    if n < 2:
        return 1
    return fib(n - 1) + fib(n - 2)


def main():
    with concurrent.futures.ProcessPoolExecutor() as executor:
        executor.map(fib, FIBS)


if __name__ == '__main__':
    main()

That should get it to work:

@DonJayamanne One thing to note is that we need to pass pathmappings in this case even though it is launch. Since we attach to the child process.

        {
            "name": "Terminal",
            "type": "python",
            "request": "launch",
            "program": "${file}",
            "console": "integratedTerminal",
            "subProcess": true,
            "pathMappings": [
                {
                    "localRoot": "${workspaceFolder}",
                    "remoteRoot": "."
                }
            ],
        },

[karthiknadig]

The above answer should resolve this issue.

[machineCYC]

@karthiknadig Thanks for this method, it can work. But have other methods without using
if name == 'main':
main()

just using
with concurrent.futures.ProcessPoolExecutor() as executor:
executor.map(fib, FIBS)

[int19h]

The key part for the workaround set_start_method - so long as you can somehow call that before any processes get spawned, it should work. If you have a single module, then doing that on top level, as in the code above, will suffice. If there are multiple modules, you can add a new module that just does set_start_method, and import it from the rest.

The __name__ == '__main___' bit is a general requirement for any module that has a function invoked via multiprocessing with the "spawn" implementation (which is the only one on Windows, so it's a requirement for portable code). But it's only needed if the module runs any code on top-level that you wouldn't want to also run in the child process when it starts:

https://docs.python.org/3/library/multiprocessing.html#the-spawn-and-forkserver-start-methods